SOC 2 for SaaS Companies: Map the Service Before the Controls
Start a SaaS SOC 2 program with a service map: confirm the buyer's report request, define the system boundary, assign control owners, and keep source evidence.

SOC 2 for a SaaS company starts with the service customers use, not a generic control checklist. Ask the buyer which report, service, Trust Services Categories, and date or period it needs. Then map the people and systems that deliver that service, assign owners to the relevant controls, and keep dated evidence from the systems where the work happens. An independent CPA firm examines the agreed scope and decides whether the evidence is sufficient.
Does your SaaS company need SOC 2 now?
A buyer request is a more useful starting point than a company-size rule. Ask for the exact wording from procurement or the contract. Does the buyer need a Type 1 or Type 2 report? Must it name your production SaaS service? Which Trust Services Categories matter to the buyer? What report date or period will it accept?
The startup SOC 2 decision guide covers the choice to start or wait. If the buyer needs a report, do not promise a date until your team and CPA firm have checked the scope, current controls, and evidence history. A report is the result of an independent examination, not the output of installing a GRC tool.
Draw the SaaS service boundary
The AICPA SOC 2 Description Criteria call for management to identify the services and describe the infrastructure, software, people, procedures, and data that make up the system. For SaaS, start with a real customer action, such as signing in, uploading a file, running a job, or exporting a result. Trace where that action goes and who can change or support each step.
| Part of the service | Question to answer | Source to inspect |
|---|---|---|
| Commitments | What has the company promised about security, availability, processing, data handling, and response? | Customer contracts, SLAs, published policies, internal requirements |
| Customer and data | What does the customer send, view, export, or delete? | Product behavior, data flow, contract terms |
| Application | Which APIs, workers, stores, and admin paths handle the action? | Deployed architecture and configuration |
| Shared systems | Which identity, source control, CI, cloud, monitoring, and support tools affect it? | Access lists, deployment settings, runbooks |
| Providers | Which outside services receive data or run part of the system? | Provider inventory, terms, service settings |
| People | Who can deploy, grant access, respond to incidents, or change customer data? | Roles, approvals, on-call records |
| Procedures | Which reviews, backups, changes, and responses must happen repeatedly? | Actual work records and schedules |
Use a data flow diagram to check the path against the deployed service. The general scope guide explains exclusions and shared boundaries. One product can rely on company-wide identity and support, while another product stays outside the described service. Name both facts clearly instead of using a product label as the whole boundary.
If staging holds a copy of customer data, the production data in staging guide helps you map the copy, access, retention, and deletion decision.
Your cloud provider is part of this map, but its SOC 2 report does not become your report. AWS’s shared responsibility guidance explains that customer duties vary with the cloud services used. For example, the provider may protect the underlying infrastructure while your team manages customer data, permissions, application code, and settings. Check the actual provider agreement and service configuration before assigning a control.
Connect controls to work that already runs
The AICPA Trust Services Criteria set criteria for the examination. They do not prescribe one SaaS architecture or a fixed list of screenshots. Write each proposed control around a real risk and a service commitment or system requirement. Then name the owner, action, cadence, and source record. For a small production SaaS service, start by checking these paths:
| Work path | Control question | Example source record |
|---|---|---|
| Access | Who can reach production and customer data, and who reviews grants? | Identity export, approval, dated access review |
| Releases | Who reviews a change and which build reached production? | Pull request, build result, deployment record |
| Monitoring | What signals show service trouble, and who responds? | Alert rule, incident or review record |
| Recovery | Can the team restore the data and service it promises to restore? | Backup settings and a completed restore test |
| Providers | Who checks a provider that affects the service? | Provider review and current service details |
These are questions to test against your service, not a universal SOC 2 control set. A SaaS team with a background processor, tenant-specific data, or a customer-facing API may need different controls and evidence from a team that operates a simple hosted dashboard. Document the real design and review it with the CPA firm.
If functions and event sources run your product, the serverless SaaS guide traces a customer request, background job, logs, and failure path to their source records.
For each control, keep the record that shows the action happened. A policy saying “we review access” does not show who reviewed the production users, when, or what they found. A successful deploy does not show that the change had the required review. Record a population and cutoff where the CPA firm may need to select samples; the audit populations guide shows how to make that list checkable.
Keep the report promise tied to the evidence period
A Type 1 examination addresses control design and implementation as of a specified date. A Type 2 examination also looks at control operation during a specified period. If your team has not run a monthly review yet, a new schedule is a plan, not a completed review. If a log source was enabled last week, it cannot show events from earlier in the period. Tell the CPA firm about those gaps while planning the report, and tell the buyer only what the issued report actually covers.
Make a small evidence index before fieldwork: control, owner, source system, period, population, evidence location, last check, and open gap. Keep source records in the systems that produce them, then retain or reference fixed, reviewable copies according to your retention needs. Do not put live secrets or customer data that may need erasure into a Git repository.
FileGRC is a Git-native GRC workspace for SOC 2 work. In the FileGRC workspace, JSON holds structured records, Markdown holds long-form work, and Git supplies the change history. A founder or engineer can connect the service boundary, control owners, recurring work, and evidence references so an agent can inspect gaps and prepare sourced audit answers. Starter records are proposals, not compliance claims. Your team operates the controls and reviews customer statements; the independent CPA firm performs the examination and judges evidence sufficiency.
Run your SOC 2 program as files in Git.
Keep policies, controls, work, and evidence indexes in a repository your team and agents can inspect. Add optional FileGRC Autopilot email and Slack reminders to keep work moving.
Frequently asked questions
Does every SaaS company need a SOC 2 report?
No universal rule makes every SaaS company obtain a SOC 2 report. Ask your buyers what assurance they need, which service and report type they expect, and when they need it. Management can then weigh that demand against scope, control work, cost, and a CPA firm's advice.
What belongs in a SaaS SOC 2 scope?
Define the service being examined, then trace the infrastructure, software, people, procedures, and data used to provide it. Include relevant shared identity, deployment, support, monitoring, and provider paths. Management drafts the system description and agrees on the examination scope with the independent CPA firm.
Does our cloud provider's SOC 2 report cover our SaaS app?
No. A provider's report covers the service and controls described in that report. Your company remains responsible for the application, customer data, access, configuration, and other duties assigned to it by the service design and provider terms. Review the provider report and your own control evidence separately.
Should a SaaS company start with SOC 2 Type 1 or Type 2?
Ask the report user which type it will accept. Type 1 concerns control design and implementation as of a specified date. Type 2 also concerns whether controls operated across a specified period. Review your current controls, evidence history, and proposed date or period with a CPA firm before promising either report.
Can FileGRC collect SaaS evidence from cloud and identity systems?
No. FileGRC does not automatically pull evidence from cloud or identity systems. Those systems operate controls and produce source evidence. Your team retains or references reviewable copies in a Git-native GRC workspace, reviews the records, and works with a CPA firm that performs the examination.